Blog · Published · Updated
Learning from completed project archives
Short answer
An AI knowledge base on the archive of closed projects lets a manager ask about a specific problem and see past lessons with the circumstances they came from, linked to the project summary. The PMO chooses the material and judges whether the answer keeps the context. At Syntalith an AI app on documents costs from €3,000 net. At 150 questions a month with 25 minutes of searching each, it pays back in 1.6 months.
How we workDetails ↓
- 01Call30 minutes with an engineer, free
- 02PrototypeA working system on your process and data, free
- 03ImplementationTarget and fixed price in the contract; 50% at signing
- 04Acceptance50% once the agreed target is met
- net per application, fixed price before contract
- from €3,000
- from signing the contract to acceptance
- 4–10 weeks
- free prototype on your documents
- €0
We belong to the OpenAI and Anthropic partner programmes and learn through their official courses. What this means ↓
Author: Syntalith · 3 min read
Statistics
Project knowledge and AI: statistics 2026
1.6 mo.
payback of a €3,000 app at 150 questions to the project archive a month: €2,181 of searching today, €330 with the app. Scenario.
49%
According to Anthropic (2024), combining contextual embeddings with BM25 search cut failed retrievals by 49%, and by 67% with reranking.
20.0%
According to Eurostat (2025), 20.0% of EU companies with at least 10 employees used AI.
Source: Eurostat, 20% of EU enterprises use AI technologies (2025) ↗
50%
According to the European Commission (2026), half of EU SMEs and mid-caps name high investment costs as a barrier to AI.
Source: European Commission, Flash Eurobarometer 578 (2026) ↗
90%
According to the European Commission (2026), 90% of companies using AI say that people take the decisions, and 2% of decisions are fully automated.
Source: European Commission, Flash Eurobarometer 578 (2026) ↗
Figures checked: 22 October 2026
What previous projects tell you about workshop preparation
A manager is planning workshops with a client and asks: „What made it hard before to collect agreements from several departments?”. One summary describes the absence of the person authorised to decide, another describes material sent too late, and in a third the workshops went to plan because the team had talked through the disputed points with the departments beforehand.
Project A
no decision-maker
The person authorised to decide was missing from the workshop.
Project B
late material
Participants received the material too late to prepare.
Project C
talks before the workshop
Disputed points were discussed with the departments in advance.
A good answer shows these lessons separately and leads to the summaries. The manager does not get a false shared cause, and the successful preparation stays the lesson of one specific team, which the manager relates to the new project. Someone who did not work on the old projects does not need to know the title of the presentation or the name of its author.
The summary has to keep the cause
If a document describes only the final status, the assistant cannot rebuild the conversations that led to it. People who know the project can add an explanation, but causes should not be written in just because they sound plausible.
The PMO separates the team's observation from a recommendation the company adopted later. „An earlier meeting helped” does not mean every project has to start that way. When the company has set a shared way of working, the answer also points to the instruction in force. NASA describes its lessons learned database as a collection of officially reviewed lessons with a description of the event and recommendations, which is a model for organising project knowledge.
What a project archive knowledge base costs and when it pays back
- Today
- 150 questions × 25 min = 62.5 h (€2,181)
- With the system, a person checks the output in 15% of the time
- €327
- Model, 1 call per case, Claude Sonnet 5.5
- €2.98
- Cost of one case
- €14.54 today, €2.20 with the system
- Saving
- €1,851 a month
Payback of €3,000
1.6 months
The volume, the time and the checking share are the scenario's assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat. Model: 8,000 input tokens and 700 output tokens per call at Anthropic's rates, converted at the ECB rate of 1.1205 USD per euro on 22 October 2026.
Retrieval decides the quality of the answer. According to Anthropic (2024), combining contextual embeddings with BM25 search cut failed retrievals by 49% (Contextual Retrieval). That matters when a question joins lessons from different projects.
When an organised archive is enough
Few summaries
a shared catalogue with lesson topics
no new app
Vague summaries
first improve what the summaries say
before AI
Questions span many projects, the archive is unfamiliar
an AI app with links
from €3,000
Start with a question about previous work
- 01ProjectsA chosen group of finished engagements.
- 02QuestionsA list of what managers look for before a start.
- 03MaterialThe PMO names the summaries and the access limits on client material.
- 04AssessmentManagers check whether the answer helps the plan.
The scope of an assistant on documents is described on the private ChatGPT for your company page. Access to client material has to keep the agreed limits, and judging whether a lesson fits a new plan stays with the manager. For the call, bring an example question and the place where the summaries sit today. We check the answers in a test on questions managers have already asked, before anyone decides to build.
Worked example · Scenario
How we count the payback: answers from company documents
We run the same count for your documents in the free analysis.
€3,723
a month for finding answers to 800 questions in documents
Payback: 1 mo.
Our assumptions are 8 minutes to find an answer in documents or ask a colleague, and checking in 15% of the old time. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat, for companies with 10 or more employees. Model cost: Claude Sonnet 5.5, about 6,000 + 500 tokens per answer from documents, at the ECB rate of 1.1205 USD per euro on 22 October 2026.
See the full calculation →
