AccAnalysisAccAnalysis
Professional Services & Staffing

Project status without chasing anyone for it

Progress, blockers and next steps were scattered across meetings and tools. We built an assistant that reads the knowledge base, attends the meetings, and answers the status question directly.

At a glance
Status
Delivered
Region
United States
Engagement model
Joint lab

Status: Delivered. Described from our own delivery records. Client identity, brand and commercial terms are withheld. Figures, where given, cover the period stated and nothing beyond it.

The organisation

Context

An organisation running multiple concurrent projects, where the current state of any one of them was distributed across a knowledge base, a set of tools, and whatever was said in the last meeting.

The brief

The problem

  • Answering “where is this” required asking several people and reconciling their answers.
  • Meetings were being held largely to transfer status, which is an expensive way to move information.
  • Decisions made in conversation were not reliably captured anywhere.
  • Written updates were inconsistent, because writing them competed with doing the work.
The work

What we built

An assistant grounded in the organisation's knowledge base

So answers cite what is actually recorded.

Automatic meeting attendance

With capture and transcription, so what is said in a meeting becomes part of the record rather than evaporating.

Daily and weekly progress summaries

Produced from that record rather than requested from people.

Direct question answering

On progress, blockers and next steps.

Method

How we delivered it

  1. 1

    Establish what the record is

    Which sources count as authoritative, and what an answer should be grounded in.

  2. 2

    Build retrieval over the knowledge base

    With citation back to source.

  3. 3

    Add meeting capture

    With explicit consent handling and clear indication that the assistant is present.

  4. 4

    Add summarisation

    Reviewed against what a good project manager would have written for the same period.

  5. 5

    Pilot on a small number of projects

    Before extending.

Sequence

How it was phased

PhaseDurationWhat happens
1Source definition
1–2 wks

Authoritative sources, grounding rules, consent policy

2Retrieval layer
3–4 wks

Knowledge-base ingestion, retrieval, citation

3Meeting capture
2–3 wks

Attendance, transcription, participant notification

4Summarisation & Q&A
3–4 wks

Daily and weekly updates, question answering

5Pilot
2–3 wks

A few projects, output compared against human updates

Indicative phasing for work of this shape. Actual duration varies with data quality, access and decision speed.

Hand-over

What the client keeps

  • The assistant and its source
  • Retrieval and grounding configuration
  • The consent and notification policy as implemented
  • The summary history
Stack
PythonLLM with retrievalDeepgramMeeting-capture APIsKnowledge-base integration
Keep reading

Related

Related case studies

Tell us what your current system can't do, and we'll tell you what it would take to change that.