CONTEXT(CLONE)
Sign in

Tender questionnaire answered from approved sources

For: Bid manager or solutions consultant responding to a large tender or security questionnaire

Pattern: Hierarchical decompositionNeeds scaleDesigned for 20 to 500 agents
Get early accessSign in to use

The pain today

A tender arrives with a very long questionnaire and a short deadline. Answers are pasted from old responses, some outdated, some contradicting each other, and nobody can say which source each answer came from.

The ask

“I attached the tender documents and questionnaire, our approved answer library, product documentation and security policies. Draft an answer to every question using only those sources, cite the source for each, and list the questions we cannot answer from what we have.”

Plain words, as you would say it to a colleague. Edit it to fit your case before you send it.

What you attach or connect

  • Tender documents and questionnaire
  • Approved answer library
  • Product documentation
  • Security and privacy policies
  • Mandatory requirements list

The unit of work

One worker task per one questionnaire question.

Why a swarm fits

A tender is a tree: lots, sections, questions. Each question needs a few passages from the library, and each section can be checked for consistency before it rolls up.

Not for

A short questionnaire of familiar questions: the bid manager answers it faster from memory.

The decision tree

6 typed decisions, each with an action for every answer

At fixed moments in a run, the engine puts one narrow question to a decision model. The decision model never writes text: it answers yes or no with a probability, picks from listed options, or gives a score, about a small slice of the material. The engine then does exactly what this tree says, which is what makes the run auditable. The thresholds are the template's design values, not measured results.

  1. Planner, while planning

    Split or keep serialYes or no, with a probability

    While planning

    Does this tender section contain questions on unrelated subjects that draw on different source documents, so they can be answered in parallel?

    Sees only: The section's question list with subject labels

    Why: Related questions stay together so their answers do not drift apart.

    • Yes: 0.60 or higherthenSplit the work
    • Unsure: 0.30 up to 0.60thenKeep serial
    • No: below 0.30thenKeep serial

  2. Split or keep serialYes or no, with a probability

    While planning

    Does this question bundle several requirements that need separately sourced answers, such as hosting location, encryption and staff vetting in one item?

    Sees only: The text of one questionnaire question

    Why: Compound questions are where an answer covers one part and silently skips the rest.

    • Yes: 0.60 or higherthenSplit the work
    • Unsure: 0.30 up to 0.60thenKeep serial
    • No: below 0.30thenKeep serial

  3. Scope checkA choice among options

    Before work starts on a unit

    Does the shortlist contain a source passage on this question's subject?

    Sees only: One question and the headings of its shortlisted passages

    Why: Unsourced questions are left blank and listed, never answered from a model's imagination.

    • An approved library entrythenAccept
    • Only product or policy documentsthenEscalate to a strong model
    • Nothing on the subjectthenMark unresolved

  4. After workers, the judge checks

    Evidence checkYes or no, with a probability

    After a worker answers

    Is every commitment in the draft answer, such as a certification, a response time or a hosting region, stated in the cited passage in equivalent terms?

    Sees only: The draft answer and its cited passage

    Why: A tender answer is a contractual statement; it may say no more than its source.

    • Yes: 0.90 or higherthenAccept
    • Unsure: 0.50 up to 0.90thenEscalate to a strong model
    • No: below 0.50thenReject and retry

  5. Reconciler, while merging

    Conflict checkYes or no, with a probability

    While reconciling

    Do these two answers in the same response give different facts about the same thing, such as two retention periods or two sets of support hours?

    Sees only: Two draft answers on the same subject

    Why: Evaluators notice contradictions, and they usually trace back to an outdated library entry.

    • Yes: 0.70 or higherthenMark unresolved
    • Unsure: 0.40 up to 0.70thenEscalate to a strong model
    • No: below 0.40thenAccept

  6. Accountable person, before anything is settled

    Person decidesA choice among options

    Before anything is reported as settled

    What kind of statement is this answer?

    Sees only: One verified answer and the mandatory requirements list

    Why: Named experts approve every commitment; routine descriptive answers do not queue behind them.

    Accountable: The bid manager and named experts approve every commitment; a tender answer is a contractual statement.

    • Commitment on a mandatory requirementthenAsk a person
    • Partial or negative compliancethenAsk a person
    • Descriptive answer from an approved entrythenAccept
    • Left blank for lack of a sourcethenAsk a person

The fleet: who does what

Model tiers by role, not brands: you choose the models. Strong reasoning models plan and reconcile, small fast models do the wide work, and the judge is a decision model from a different family, so it does not share the workers' blind spots.

  1. Planner

    A strong reasoning model breaks the tender into sections and questions and shortlists sources for each.

    Decisions here:1. Split or keep serial2. Split or keep serial3. Scope check

  2. Workers

    Small fast workers from an open-weight family each draft one answer from its shortlisted passages, with citations.

    Designed for 20 to 500 agents, one worker task per one questionnaire question. Each worker receives only its own unit.

  3. Judge, from a different model family

    A decision model from a different family checks each answer says no more than its cited source.

    Decisions here:4. Evidence check

  4. Reconciler

    A strong reasoning model checks each section for contradictions, then the whole response, before rolling up.

    Decisions here:5. Conflict check

  5. Accountable person

    The bid manager and named experts approve every commitment; a tender answer is a contractual statement.

    Decisions here:6. Person decides

Checked before anything is accepted

  • Every answer cites the library entry or document passage it uses
  • Answers with no supporting source are left blank and listed
  • Answers within a section are checked against each other for contradictions
  • Each mandatory requirement is traced to an explicit yes, no or partial

What comes back

  • Draft answers with citations
  • Unanswerable questions routed to named experts
  • Contradictions between answers or sources
  • Mandatory requirement compliance table
  • Library entries that look outdated

What to measure

  • Answers accepted by subject experts without edits
  • Contradictions caught before submission
  • Bid team hours per tender
  • Questions left unanswered

Names of measures only. No result is claimed for this template.

Templates open in the workspace chat with the ask filled in. Nothing runs until you send it.

Get early accessSign in to use

More in Sales and marketing

A quarter of sales calls mapped to objections

For: Sales enablement lead or revenue operations manager explaining why deals are lost

The loss reason in the pipeline is a dropdown picked by the seller.

Pattern: Map, verify, reduceNeeds scale5 decisionsDesigned for 40 to 800 agents

Competitor pricing and product page watch

For: Product marketing manager or competitive intelligence lead keeping battlecards current

Battlecards are accurate the week they are written.

Pattern: WatchtowerNeeds a connector5 decisionsDesigned for 10 to 300 agents