Skip to content
60x
  • Company Brain
  • Knowledge Graph
  • Industries
  • About
  • Case Studies
  • Blog
Book a demo →Book a demo →
All posts
Private Equity•18 Nov 2026•12 min read•Fergus McKenzie-Wilson

Signed, not live

Signed revenue waits in onboarding. Much of the wait is people rediscovering what past projects already settled.

Key takeaways

  • Complex onboarding often stalls while teams rediscover how previous projects handled similar deviations.
  • A company brain can connect past scopes, configuration decisions and delivery records to the next customer's requirements.
  • Measure knowledge work separately from calendar delays such as approvals, bank processes and testing.

On this page

  1. The standard build is fast. The deviations take the months.
  2. Six things that stretch a go-live into months
  3. What onboarding looks like from the inside
  4. Every deviation starts a loop
  5. Where the time goes
  6. A company brain over the delivery knowledge
  7. The same idea, already running
  8. Agents for the point and click
  9. What changes, and what stays the same
  10. Why investors care
  11. How to test it on one company
  12. Sources
  13. Frequently asked questions
  14. Related reading

I recently spoke to the AI team at a large software investor. They didn't want help finding deals. They wanted help with something that happens after a deal is won. A company brain can connect the past delivery evidence.

"A customer could easily take six or nine months to onboard to your product," one of them told me, describing complex treasury software sold to banks and large enterprises. "A lot of the steps that we're doing to onboard a customer are like either pre-sales information or point and click in the software and configuration."

His colleague put the cost in one line: "Companies that have big backlogs are trying to pull forward the revenue." Then he named the hard part: "the deviations from standard. What are the clusters of companies that have similar deviations?" We see a related problem when sales promises and delivery records drift apart.

That question is the subject of this post. Signed revenue waits in onboarding. Much of the wait is people rediscovering knowledge that already exists inside the vendor, spread across past projects.

Signed, Not Live
Signed revenue waiting
Signed: ARR booked
OnboardingDiscoveryRule setupBank linksChange controlTesting
Revenue starts
0123456789

Months from contract signed

“A customer could easily take six or nine months to onboard.”

AI lead, software investor

Illustrative. Six to nine months is the onboarding time an investor described for complex treasury software.

The Deviations

The standard build is fast. The deviations take the months.

We looked at enterprise finance-automation platforms that sit inside SAP: cash application, accounts payable, payments and collections. The vendors in this category market fast go-lives, and for the standard part of the setup that holds. Most of a customer's configuration matches the product's defaults, and an experienced consultant can set it up in days.

The months go on the rest. A bank sends its own file variant. An approval matrix differs by country. A posting rule was set years ago in a legacy tool and nobody remembers why. Consultants find these deviations project by project. Services make up a large share of revenue at many of these vendors, and delivery depends on a small bench of senior consultants who carry the deviation knowledge in their heads.

The scale of the customers explains the calendar. Published case studies in this category describe single rollouts covering:

  • dozens of countries on one template
  • hundreds of legal entities
  • more than a hundred banks, each with its own formats and signatories

Regulation adds more onboarding on top. France's e-invoicing mandate took effect on 1 September 2026: every business must be able to receive electronic invoices, and large and mid-sized firms must issue them. Germany's mandate is phasing in from 2025 to 2028, Spain's first wave is planned, and the EU's ViDA package follows in 2030. For a finance-software vendor, each mandate means a wave of customers to reconfigure at once.

Grant Watling and Dhruv Sethi, who work across Hg's portfolio on AI products and operations, described the pressure in September: "Implementation tolerance has shrunk from weeks to days." They report that one portfolio company halved its implementation time while cutting delivery costs by 30 percent.

Root Causes

Six things that stretch a go-live into months

The causes are consistent across the vendors we looked at.

  • Knowledge in heads. Which bank needs which format, and why a rule was set, sits with a few senior consultants.
  • SAP change control. Every change to an SAP add-on goes through the customer's transport and change-management process.
  • Bank by bank. Each host-to-host, EBICS or SWIFT link has its own onboarding and test cycle.
  • Forced migrations. Legacy on-premise SAP add-ons can't run on S/4HANA Public Cloud, so customers have to move the base.
  • Acquired products. Customers of acquired products sit on separate stacks, with rules nobody still at the company documented.
  • Late deviations. Non-standard behaviour found in testing becomes change orders and slipped waves.

Some of these are calendar time. A bank's test cycle takes as long as the bank takes. Others are knowledge time: people finding, re-reading and re-deciding things that an earlier project already settled.

From The Inside

What onboarding looks like from the inside

We work with an AI accounts-payable company that sells to large enterprises on SAP. This autumn it onboarded several large industrial and automotive groups, and each followed the same sequence:

  1. An introduction, then an IT and security deep-dive.
  2. Four to six weeks of discovery, with seven or eight deep-dive sessions per customer, one for each process in scope.
  3. An on-site workshop.
  4. A twelve-week proof of concept, with baselines taken from the customer's own data.

The same blockers came up at every customer, and none of them was technical:

  • NDAs and data-processing agreements, before any data could move.
  • Finding the customer's internal SAP owner.
  • Getting an SAP add-on transport installed in the customer's QA system.
  • Mailbox access, which needed sign-off from the customer's identity team.
  • Moving files at all: several of the customers block common file-sharing tools.

The same knowledge problems came up too:

  • routing rules kept in Excel lookup tables
  • prioritisation that lived in team members' heads
  • no written process for supplier reconciliation
  • staff-to-branch mappings missing for part of the workforce

At one customer, the person who assigned invoices to projects had two to three years of experience and no documentation. Asked whether they could explain the logic in a three-hour interview, the team said no.

After every walkthrough, the vendor wrote up the customer's process for the customer to check, then a solution proposal. Across seven or eight sessions per customer, that is the most repeated piece of writing in the whole onboarding.

Deviation Loops

Every deviation starts a loop

The deviations we saw were concrete enough to count. At one customer's bank-payment posting, the standard rule matched each incoming payment to a customer and posted it. The deviations were:

  • The bank leaves out data. Some banks send an incomplete tax ID, so SAP can't match the payer.
  • The payer isn't the customer. A relative or a business partner pays on the customer's behalf. Posting then needs a separate transfer approval.
  • The customer number never appears on the statement. It comes from a sales adviser by email, chat or phone.
  • The posting codes carry meaning. They encode branch and product line together, so a payment posted with the default code lands in the wrong profit centre.
  • Some banks use transaction codes the import rules don't know. Those payments never reach SAP until someone files an IT ticket.

On accounts payable, the same company catalogued about 50 types of exception across the invoice lifecycle: supplier records, intake, routing, tax coding, matching, approval, posting and payment. Its sharpest finding: an exception is a loop. At one customer, a VAT code on an invoice that disagreed with the purchase order forced the order to be reopened and re-approved, because released documents couldn't be edited. One wrong two-digit code re-ran an approval chain of three to five people.

Each of these varies by customer, along a short list of dimensions:

  • company codes, and which shared service centre pays for which entities
  • separate payment terms for finance and for purchasing
  • the share of invoices with and without purchase orders
  • blocking reasons: price, quantity, date and budget
  • bank statement formats
  • country VAT rules
  • the number of SAP instances, and which company codes sit on each

That is good news for anyone onboarding the next customer. If the dimensions are known and countable, a new customer can be matched to the past customers it most resembles.

A Deviation Is A Loop
Purchase order released
Invoice arrives
Match
VAT code disagrees with PO
Approve (3 to 5 people)
Post

Released documents can't be edited: reopen the PO.

Fixed once, remembered: the brain knows this customer's VAT rule before the first invoice arrives.
Illustrative, based on an accounts-payable process we have seen. One wrong two-digit code re-ran the approval chain.

Where The Time Goes

Where the time goes

Split a typical enterprise rollout by phase and the pattern is clear.

Enterprise rollout phases
Phase Knowledge work or calendar time?
Sales handover and scoping Knowledge
Blueprint and fit-gap Knowledge
Rule and deviation setup Knowledge
Data and rule migration Knowledge
Bank links and formats Calendar
Install and change control Calendar
Testing and go-live Calendar

The first four phases are where a consultant asks the customer questions the vendor has heard many times before. Somewhere in a past statement of work, a fit-gap document, a change order or a ticket, the answer exists for a customer that looked like this one. Nobody can find it in time, so the team runs the discovery session again.

The deviations cluster by process:

Typical deviations by process
Process What gets set up A typical deviation
Cash application Posting and matching rules by bank, customer and entity Remittances with no invoice reference
Bank connectivity Statement and payment channels per bank A bank's own file variant or test cycle
Accounts payable Matching tolerances, approval matrices, coding Non-PO invoices and per-country VAT rules
Collections Dunning levels, segments, credit limits Disputes and deductions by customer group
Payments Approval chains, signatories, run schedules A payment factory across hundreds of entities

These patterns repeat. A US subsidiary on a lockbox, netting across entities, a bank that sends a CAMT.053 variant: a vendor that has run 200 implementations has seen each of these many times.

Knowledge Time And Calendar Time
Knowledge: finding and re-deciding what past projects settledCalendar: waiting on banks, IT and the customer
Sales handover and scoping
Blueprint and fit-gap
Rule and deviation setup
Data and rule migration
Bank links and formats
Install and change control
Testing and go-live
Illustrative phases of an enterprise rollout.

Company Brain

A company brain over the delivery knowledge

A company brain lets an onboarding team ask anything the vendor has already learned, and get an answer from its whole delivery history:

  • statements of work, blueprints and fit-gap documents
  • posting, matching and approval rule exports
  • bank lists, with their channels and formats (EBICS, SWIFT, MT940, CAMT.053)
  • change orders and support tickets
  • CRM records, RFP answers and contracts
  • notes and transcripts from kick-offs and process walkthroughs

For any past customer, the team can see its entities, ERPs, banks, rules, deviations and outcomes in one place. Every answer cites the document, ticket or call it came from. The data stays in the vendor's own tenant, and people only see what their access to the source allows.

That gives the team four things aimed at the slow work.

  1. An onboarding brief before kick-off. For each signed customer, the brain drafts what the team would otherwise spend the first weeks collecting: entities, ERPs, banks, formats, likely deviations, open risks and contacts. Kick-off then confirms and corrects a draft instead of starting from zero. Gaps the brain can't fill become the data request.
  2. Deviation clusters during design. The team can see which past customers share a deviation. A new customer starts from the closest configuration that has already worked, instead of a blank template. This is the investor's question in practice: which companies deviate in the same way?
  3. A migration accelerator. For customers moving off legacy on-premise add-ons or acquired products, existing rule sets arrive already drafted in the cloud platform's setup. A consultant reviews every translation.
  4. Walkthrough to proposal. Each process walkthrough becomes a written summary and a solution proposal, ready for the customer to check, with links back to the call.

Here is what a brief looks like, with mock data:

Onboarding brief (mock data)
Item Draft from the brain
Customer Industrial group, 40 entities
ERPs SAP S/4HANA (EU), legacy ECC (US)
Banks 14 banks via EBICS; CAMT.053 and BAI2
Legacy to retire SAP VIM and an OCR tool
Likely deviations Netting across entities; lockbox in the US
Nearest past customers Customer A, Customer B
Open risks Change-control transports on ECC

Nobody has to write the process down first. Some software companies have detailed runbooks, and others run delivery out of people's heads. The second group gets the same answers from what it already has: tickets, configuration exports, call notes and email.

From Delivery History To Onboarding Brief
What goes in
SOWs and fit-gap documents
Rule exports
Bank lists and formats
Change orders and tickets
CRM, RFPs and contracts
Kick-off and walkthrough notes
Company brain
CustomerEntityERPBankRuleDeviation
In your tenant
What comes out
Onboarding brief
ERPs
SAP S/4HANA (EU), legacy ECC (US)
src
Banks
14 banks via EBICS; CAMT.053 and BAI2
src
Likely deviations
Netting across entities; lockbox in the US
src
Deviation clusters
Migration drafts
Walkthrough to proposal
Illustrative. The brief uses mock data.

Already Running

The same idea, already running

We use this approach at an executive search and advisory firm. Its company brain covers about 10,000 SharePoint files and has been live since spring 2026. The firm-wide version we are now building lets consultants find the precedents across 20 years of client mandates: past mandates that share a capability, a geography, a growth goal or an outcome.

That is the deviation-cluster idea in a different industry. A new mandate starts from the closest past one, with the evidence attached.

The same build answers the sensitivity question. The firm knows which people and passages can leave the firm. Client-confidential material can inform an analysis without appearing in it. A software vendor's customer configurations and bank details need the same treatment.

Agents

Agents for the point and click

The investor's second point was that much of onboarding is "point and click in the software and configuration." Once the knowledge is in one place, an agent can do much of that clicking, and consultants spend their time reviewing instead.

The loop has four steps:

  1. Ask. At each admin screen, a computer-use agent asks the brain which rule, bank or deviation applies to this customer.
  2. Configure. It works through the product's admin screens and SAP GUI: posting rules, tolerances, approval flows and bank formats.
  3. Test. It replays sample statements and invoices from similar past customers and checks the results.
  4. Approve. A consultant reviews the logged changes and approves them. Each outcome is written back to the brain, so the next customer starts from it.

Vendors in this space are converging on the same order: context first, then agents that act on it. The AP company above drafts written procedures for each customer from its SOPs, screen recordings and interviews. It is also working on extracting configuration and history straight out of SAP to shorten onboarding. An agent that clicks through admin screens without that context reproduces the template. With it, the agent can reproduce the customer's deviations.

The agents work in a test or QA system and never in production. The consultant promotes approved changes through the customer's own change control.

The Configuration Loop
1. Ask the brain
2. Configure in QA
4. Consultant approves
3. Replay past cases
BrainCustomers known120
Outcome written back after step 4Promoted through the customer's change control
Production
Agents configure in a test or QA system. A consultant approves every change before it reaches production.

What Changes

What changes, and what stays the same

Knowledge time shrinks. Calendar time stays.

  • Discovery. Weeks of deep-dive sessions become days spent confirming a sourced draft.
  • Configuration. Starting from the closest proven setup means fewer deviations turn up late in testing.
  • Bench capacity. Each senior consultant can carry more projects, because less of the project depends on what they remember.

Several things don't change:

  • the NDA and the customer's security review
  • installing add-ons under the customer's change control
  • bank connectivity and format testing, one bank at a time
  • user acceptance testing, and time from the customer's own experts

Data stays in the vendor's tenant, and no customer data trains a model.

We won't quote a percentage saving before a pilot measures one. The size of the gain depends on how repetitive a vendor's deviations are and how much of its history is written down.

Today Against With A Brain
Today
Discovery, design, deviations
Banks, install, testing
With a brain
Confirm draft
Banks, install, testing
What a pilot measures
Illustrative. The size of the gain is what a pilot measures; we won't quote a number before then.

For Investors

Why investors care

For a software investor, onboarding time shows up in four places in the value-creation plan.

  • Revenue. Fewer weeks from signature to go-live bring subscription and services revenue forward.
  • Margin. Less dependence on scarce senior consultants as volume grows.
  • Retention. Smoother migrations for legacy and acquired-product customers protect the installed base.
  • Exit. A before-and-after delivery metric gives buyers an AI story they can check in diligence.

Pilot

How to test it on one company

Pick one portfolio company where signed customers wait longest to go live. Then run three steps.

  1. Week 1, connect. Start from five to ten past implementation packs, plus CRM and ticket history, kept in the company's own tenant.
  2. Week 2, brief the backlog. Produce onboarding briefs for 10 to 20 signed customers waiting to start, and deviation clusters across the installed base.
  3. Then, measure. Score the briefs against a golden set of past projects the team knows well, then track the next cohort of go-lives.
Pilot metrics
Metric Baseline Read-out
Signature to go-live From recent projects Next cohort
Consultant days per project From timesheets Next cohort
Brief accuracy Golden set End of week 2

You choose where it runs:

  • 60x Cloud, with EU or US data residency
  • Your Cloud, inside your own Azure, AWS or GCP cloud environment

Book a demo and bring one portfolio company with a backlog. We'll show you a brief.

Sources

Sources

  • Grant Watling and Dhruv Sethi, The quiet reinvention of tech implementation, Hg, September 2026.
  • French Ministry of the Economy, La facturation électronique pour les entreprises.

Investor quotes are anonymised. The advisory-firm and accounts-payable examples are current 60x clients and their customers, described without names. The onboarding brief uses mock data. 60x installs company brains for investment firms and their portfolio companies.

Frequently asked questions

Why does enterprise software onboarding take so long?

Customers often need deviations from the standard product. Teams must find past decisions, agree requirements, configure the system and complete approvals and testing. Reconstructing knowledge from previous projects adds to that delay.

How can AI reduce software onboarding time?

A company brain can find comparable past projects and the evidence behind their configuration choices. Agents can help with repeatable configuration tasks, with people reviewing the output and approving changes.

How do you measure an AI onboarding pilot?

Choose one company with a backlog and compare a repeatable onboarding task before and after the pilot. Track time spent finding answers, configuration effort and output quality separately from external approval and testing delays.

Related reading

  • Sold, scoped, shipped

    Why forward-deployed teams drift out of sync with sales and engineering, and how a company brain keeps them aligned.

  • Eight companies, one brain

    How to make a roll-up's knowledge usable before the IT migration. Meet Tarnovell, a fictional group built to test it.

  • Faster alone, slower together

    The alignment problem AI made worse, and how a company brain closes the information gap.

Author

Fergus McKenzie-Wilson

Fergus McKenzie-Wilson

Founder

Pioneered Multi-Agent AI in 2023, transforming a consultancy into AI-native SaaS. Serial entrepreneur and exited founder; co-founded Hydra Drones at university and exited after 4 years. Awarded an Innovate UK grant and received multiple innovation awards from the Austrian government. Fergus architects and leads the engineering of 60x's agentic infrastructure and proprietary Knowledge Graph design.

Want help turning a research thesis into a working system?

We build the software, workflows, and knowledge layers that make AI useful inside real operating teams.

Book a demo →Book a demo →
60x

We help enterprise leaders deliver AI outcomes.

Company BrainKnowledge GraphWhat We KnewPermissionsTrust
Private EquityFinancial ServicesConsultancies
AboutCase StudiesJobs
BlogContactTermsPrivacyCompanies House

Solutions

Company BrainKnowledge GraphWhat We KnewPermissionsTrust

Industries

Private EquityFinancial ServicesConsultancies

Company

AboutCase StudiesJobs

Resources

BlogContactTermsPrivacyCompanies House
Cyber Essentials Certified
Unicorn Mafia

© 2026 60x.ai. All rights reserved.

PrivacyTerms of Use