You bought the company. Make sure you also get what the founder knows.
Knowledge Engine indexes an acquired business’s files, mail, chats, ERP and CRM into one private knowledge base, so the new management team gets cited answers about customers, suppliers and processes while the founder is still there to fill the gaps. Each portfolio company keeps its own box and data, run by Olyra as a managed service.
What is Knowledge Engine for private equity and acquirers? It is a private, permission-aware knowledge base that Olyra installs inside an acquired or portfolio company, on a box at its site or a server in its own cloud account. It reads what the business already has — file servers, mailboxes, Microsoft 365 or Google Workspace, the ERP and the CRM — and lets the new CEO, the operating partner and the staff ask questions in plain language. Every answer cites the exact source passage and is limited to what the person asking may see. During the founder’s transition, procedures and the reasons behind decisions are captured as cited, approved answers. The data and the keys stay with the company. Olyra operates the system as a managed service — updates, security patching, monitoring and support — and holds no copy of the data.
Key-person risk is in the deal model. Then it shows up in the business.
Acquirers of owner-led companies describe the same first year. The knowledge the deal depended on sits in one person’s head and inbox, and the clock on that person’s transition started at close.
“Ask the founder” is the operating system
Why this customer gets different terms, which supplier can be pushed, what was agreed on a handshake years ago. The transition period is spent answering the same questions one meeting at a time, and it ends on a date.
The 100-day plan starts without a map
New management inherits file servers, mailboxes, an ERP and a CRM, and nobody can say where the current price list, the signed contract or the real procedure lives.
People leave after a change of owner
Turnover after an acquisition is normal. When a long-serving planner, buyer or account manager goes, the exceptions and the customer history they carried go too.
Every add-on brings its own mess
Each bolt-on arrives with its own file shares, ERP and habits. Integration slows down because nobody can compare how two businesses actually handle the same thing.
Governance without a shared cloud
The fund wants audit trails, access reviews and consistent reporting. The company’s customers, lenders and counsel do not want its contracts and mail indexed in a vendor’s multi-company service.
Operating partners need answers, not everything
A portfolio-operations team has to ask questions inside a company without seeing HR files, legal matters or anything else outside its remit. Old shared-folder permissions do not make that distinction.
From “ask the founder” to “ask the business”
The knowledge is captured once, at the start of the hold period, and stays with the company after the people who held it have moved on.
- The new CEO can ask a question on day 3 and get an answer that opens the contract, e-mail or ERP record behind it
- The founder’s time goes to the gaps. What is already written down is found by the engine, so transition meetings cover the reasons and exceptions that were never written anywhere
- Those reasons are kept. A steward approves them as curated answers, which take precedence for the questions that matter most
- Procedures become explicit. SOPs and policies are read as versioned, cited steps with roles, thresholds and approval points
- Old versions stop causing mistakes. Each document is marked verified, stale, superseded or disputed, and answers say when they lean on an old one
- Each company stays separate. Its box, its data, its keys, its audit trail
This is the same product described on the main Knowledge Engine page, set up for the situation an acquirer is in: a business to understand quickly, a founder on a fixed timeline, and a governance standard to meet.
What it does for a newly acquired business
Each capability is here because it answers a specific problem acquirers run into.
A map of what exists
For the inherited file servers, mailboxes, ERP and CRM
- Reads documents, mail, chats, spreadsheets and ERP and CRM records, including scanned PDFs through OCR in the languages your documents use
- Organizes everything by categories set for that business: client, department, sensitivity, document type
- Merges different spellings of the same customer or supplier into one entity
- Shows ERP customers, orders and invoices next to the e-mails and contracts about them
Cited answers for new management
For a 100-day plan that needs facts
- Plain-language questions in English, French, Spanish, German, Portuguese and more
- Citations point to document, version and exact location
- Conflicting numbers, dates or revisions between sources are flagged, not hidden
- When the evidence is not there, it says so instead of guessing
The founder’s ‘why’, captured
For a transition that ends on a date
- Unanswered questions become a steward’s to-do list, which makes a practical agenda for founder sessions
- Answers approved by a steward are stored as curated answers and take precedence
- A “who owns this?” hint routes a question to the owning role
- Corrections staff make to earlier work become precedents, once a steward approves them
Process as it is written, and as it runs
For standardizing after close
- SOPs and policies become versioned, cited steps with roles, systems, thresholds and approval points
- Divergence findings show where tickets or mail suggest a process is running without its approval step
- Findings report documents and counts, never individual people
Permission-aware by design
For operating partners and outside advisers
- Per-person permissions are enforced inside the database, so no screen or AI assistant can go around them
- Existing file-share and SharePoint restrictions can be carried over
- A permission-hygiene check runs before go-live, so old oversharing is found first
- Evaluative questions about individual employees are refused
Governance in the box
For the fund’s reporting and audit standard
- Tamper-evident audit trail of sign-ins, searches, views, assistant calls and exports
- Retention rules, access reviews, eDiscovery with legal holds, and data-subject requests
- Two-person approval for sensitive settings, with rollback
- The whole knowledge base can be viewed as it stood on any past date
Figures that match the ERP. Named business figures, such as open receivables or orders by month, are computed from ERP records through definitions the company’s steward manages, with the query shown as provenance. Where a fund wants one trusted set of numbers per company, this sits alongside Olyra’s Operational Intelligence work.
How a deal team uses it across the hold period
The work is front-loaded on purpose. The knowledge base is most valuable when it is built while the founder and the long-serving staff are still in the building.
Get ready, in general terms
See a demonstration box on sample data, agree which source to connect first, and name a sponsor and a data steward for the company. Knowledge Engine is installed in a business with that business’s authorization. It is not a way to read a target’s confidential information before the proper agreements are in place.
Understand how the business runs
Install, connect the richest source first, and put the management team’s real questions to it. Use the unanswered ones to plan founder sessions. Add sources, set permissions and residency rules, and agree the first business metrics.
Run, govern and add on
The knowledge base stays current as documents change. Access reviews, retention and the audit trail support the governance standard. An add-on acquisition is ingested onto the platform company’s box, or gets its own box where it must stay separate.
Hand over an organized business
The knowledge base belongs to the portfolio company and its contents can be exported at any time. A future owner’s team meets a business whose contracts, procedures and decisions can be found and checked.
Knowledge Engine does not replace legal, financial, tax or commercial due diligence, and it makes no promise about valuation, deal terms or the outcome of a sale. Nothing on this page is legal, tax or financial advice.
Cited answers on day 3, inside the first week of ownership
First ingestion is real work: the data is read, extracted, classified and built into a fast database in several formats, which takes time and AI processing. So each company starts with one high-value source, not everything at once.
Install and connect one source
Sign-in is federated with the company’s identity provider. The single richest source is connected first, often the main file server or the founder’s and sales team’s mail.
The new team’s questions, answered
Ten real questions from the CEO or operating partner, asked and answered in front of them, with citations that open the source.
More sources, first assistant
A starter taxonomy is in place, the ERP or CRM is connected, and a first AI assistant does one narrow job end to end.
The first evaluation report
A golden set of the company’s real questions, measured results, first business metrics and a trained data steward.
Remaining sources, classification review, residency rules, roll-out to users and go-live sign-off follow over the next four to six weeks. What we need from each company: a sponsor such as the CEO, CFO or IT lead, a data steward with a few hours a week for the first two months, read access to the sources, and somewhere to put the machine or a cloud account.
The typical acquirer approach, and the same job with Knowledge Engine
The usual methods work, up to a point. They depend on people’s time and memory, and they leave little behind.
| The job | Typical acquirer approach | With Knowledge Engine |
|---|---|---|
| Transfer the founder’s knowledge | Weekly meetings, shadowing and a handover document written near the end of the transition | The written record is indexed first; founder sessions target the unanswered questions, and approved answers are kept with citations |
| Build the 100-day picture | Interviews, data requests to staff and spreadsheets assembled by the deal team | Management asks directly and opens the source behind each answer; conflicts between sources are flagged |
| Find the current contract, price list or procedure | Ask whoever has been there longest | Search by meaning and keyword, with each document marked verified, stale, superseded or disputed |
| Document processes | A consultant interviews staff and writes SOPs that age from the day they are delivered | Existing SOPs and policies are read as versioned, cited steps; divergence findings show where practice differs |
| Give an operating partner visibility | Broad folder access, or reports prepared by hand | A role with defined permissions, enforced in the database, with every query logged |
| Meet the fund’s governance standard | Policies on paper and manual evidence gathering when someone asks | Audit trail, access reviews, retention rules and eDiscovery produced by the system |
| Integrate an add-on | Migrate file shares and wait for an ERP consolidation | Ingest the add-on’s sources as they are, next to the platform company’s, and search across both |
| Survive staff turnover | Exit interviews and a hope that the inbox is kept | The mail, documents and decisions are already indexed and stay searchable under the company’s retention rules |
Other tools have their place. A copilot bundled with an office suite works well inside that suite and follows whatever permissions already exist. A cloud enterprise-search platform offers broad connectors, with the content indexed outside the company. A virtual data room is the right tool for a transaction, and it is not built for running the business afterwards. The main Knowledge Engine comparison covers these in more detail.
One box per company. Its data under its own control, run as a managed service.
A portfolio is a set of separate legal entities with separate customers, lenders and confidentiality obligations. The knowledge base follows that structure.
| The portfolio company’s | Olyra’s ongoing service |
|---|---|
| Its data, its knowledge base and every export, at any time | Setup: installs and hardens the system, connects the sources, and builds the taxonomy and residency rules with the company’s data steward |
| The recovery keys, handed over at go-live | Runs updates in the company’s maintenance window, with automatic rollback if a health check fails |
| The box, or the server in its own cloud account, where it runs | Monitors and security-patches the system, and supports it under an agreement with published response times |
| The decision to approve every support session and every sensitive change | Adds connectors and improvements and delivers evaluation reports on the company’s own questions |
| Its own audit trail and the evidence it produces | Holds no copy of the data; support engineers see content only with the company’s approval |
No shared index across the portfolio
Company A’s contracts are not in the same system as Company B’s. An operating partner who works with both has a separate, permissioned account in each.
AI models at addresses the company controls
Models run on the company’s own GPU, on a cloud service under its own account, or on the box. Olyra does not host AI inference, so no Olyra service sits in the path of the data.
Residency rules for cross-border portfolios
For holdings in Brazil and the US, rules decide what may leave by where content came from or what it is. A simulator shows what a rule would allow before it is switched on. If in doubt, the engine blocks and logs the decision.
Security controls are designed to map to SOC 2 and ISO 27001 expectations, and the engine produces the evidence an auditor asks for. Control mapping is not certification, and legal sign-off stays with each company’s counsel. More detail is in the Trust Centre.
Sized for one company, with room for add-ons and more sites
Each company starts as a single box. Machines are added when the business grows, with nothing reinstalled or reloaded.
Several offices after a roll-up. One site is the master and each other location holds a replica that serves its own people locally. When a replica site has local documents to add, it sends them to the master; the master reads and organizes them and replicates them back. How much each stage adds depends on hardware, data volume and configuration, and we size it with each company.
On speed. Because the data is prepared in advance, answers come from what is already built and are fast enough to use in the middle of a task. Lookups can return in as little as a couple of milliseconds. Results vary with hardware, data volume and query type.
Who this is for, and who it is not for
A good fit
- Private equity firms, search funds, holding companies and strategic acquirers buying owner-led businesses of roughly 50–500 people in Brazil or the US
- Deals where the founder stays for a defined transition and then leaves
- New CEOs and operating partners who need to understand customers, suppliers and processes quickly
- Buy-and-build strategies where each add-on brings its own file shares and ERP
- Portfolios that need audit trails and access control without placing company data in a shared vendor cloud
Not a fit
Knowledge Engine is not a virtual data room and not a diligence service. It does not replace legal, financial, tax or commercial due diligence. It is not for reading a target’s confidential data before the right agreements exist, and it is not a single multi-company cloud where a fund pools every holding’s documents. It will not evaluate individual employees: those questions are refused by design. And if a business has only a few hundred documents and no ERP to speak of, a lighter tool will do the job, and we will say so.
Questions acquirers and operating partners ask
Can we use it during due diligence on a target?
Only in general terms. Before close you can see a demonstration box on sample data and plan the first source to connect. Installing Knowledge Engine means reading a company’s files, mail and ERP, which requires that company’s authorization and the proper agreements. It does not replace legal, financial, tax or commercial due diligence, and nothing here is legal, tax or financial advice.
Can we try it on a portfolio company’s data first?
We show a demonstration box running on sample data, so you can see search, cited answers, permissions and the audit trail working. We do not set up a free trial on a company’s own data: bringing data in means reading, extracting and classifying it and building a fast database in several formats, which takes real time and AI processing. That is why each engagement starts with one high-value source and shows cited answers to the team’s own questions on day 3.
Can the fund run one system for the whole portfolio?
No, and that is deliberate. Each portfolio company has its own box, its own data and its own keys, so one company’s contracts and mail are never indexed alongside another’s. An operating partner can hold a permissioned account in each company. The same setup, taxonomy templates and governance standard can be repeated across holdings.
What can an operating partner see?
Whatever the company grants that role, and nothing else. Permissions are enforced inside the database, so they apply to the console and to any AI assistant the partner uses. Every search, view and assistant call is recorded in the tamper-evident audit trail, and access reviews show who can see what, and why.
Will the founder and the staff accept it?
It helps that the engine tracks documents and never scores people. Evaluative questions about an individual employee are refused and small-group results are suppressed, so it cannot be turned into a monitoring tool. For a founder, it replaces repeated questions with a short list of gaps only they can fill, and their answers are kept with their name on the approval.
How is this different from a virtual data room?
A data room is a curated set of documents assembled for a transaction and closed afterwards. Knowledge Engine indexes the live business, including mail, chats and ERP and CRM records, keeps itself current, and answers questions with citations for the people running the company.
What happens to the knowledge base when we sell the company?
The data, the keys and the box belong to the portfolio company, so the knowledge base is part of the business. It supports a cleaner handover to the next management team. The managed service that keeps it updated, secure and supported is with the company; how it continues under a new owner is set out in the agreement. The contents can be exported at any time. We make no promise about valuation or sale outcome.
How do add-on acquisitions fit?
An add-on’s file shares, mailboxes and ERP are connected as new sources on the platform company’s box, each labelled by origin, so staff can search across both businesses under their permissions. Where an add-on has to stay separate, it gets its own box. Its first ingestion takes time and processing like any other, so it also starts with one high-value source.
Where do the AI models run, and does Olyra see the data?
Models are reached at addresses the company controls: its own GPU, a cloud service under its own account, or the box. Olyra does not host AI inference and holds no copy of the data. No one has standing access to the box; an Olyra engineer requests a support session with a reason and a duration, and the company approves it, can watch it, and can end it.
Start while the founder is still in the building
The transition period is the one window in which this knowledge can be captured from the person who holds it. A briefing takes thirty minutes: we show the demonstration box, talk through one company in your portfolio or pipeline, and agree which source would answer the most questions first.
Book a portfolio knowledge briefing
Prefer to read first? The full product description is on the Knowledge Engine page.
Find out what AI can actually do for your operations.
Start with an AI Operations Audit — a fixed-scope diagnostic that maps your workflows, scores your AI readiness, and hands you a prioritized 120-day roadmap. If we don't find real opportunities, you'll know that too.
Request an AI Operations Audit → Prefer to talk first? Book a 30-minute scoping call.