Seventy public tenders, screened and annexed by an agent pillar
Context
South African public infrastructure procurement publishes continuously. For a specialist consultancy the constraint is never finding tenders. It is deciding, quickly and honestly, which handful are worth the cost of bidding, and then producing the technical annexes that a compliant submission demands.
Both halves of that problem consume senior time, which is the scarcest resource in a founder-led firm.
The problem
A single eTenders sweep routinely surfaces dozens of opportunities. Most are off-domain and can be discarded in seconds by a human who knows the business. The remainder need real judgement: scope fit, cost and effort to bid, delivery capacity, expected return, mandatory registrations, and closing dates that are often days away.
Traditionally that triage is a senior person reading bid documents. The annexes that follow, contracts methodology, risk methodology, market backdrop, competitor positioning, are each a specialist writing exercise that starts from a blank page every time.
The approach
The work was split across the Strategy and Intelligence pillar, with Frigg dispatching and reviewing.
Geri, the business development agent, ran the sweep and qualification: 70 opportunities screened against a fixed tier definition. Tier A core fit, Tier B panels and frameworks, Tier C feasibility and advisory, and everything else discarded as off-domain. That produced 28 qualified entries, of which 9 were Tier A.
Five specialist agents then produced the reusable annexes: Vor built an incumbent and competitor map of the firms likely to bid, Gefjon produced a municipal water and sanitation market read with the grant instruments and fiscal envelope named and sourced, Forseti built a contracts technical approach spanning FIDIC, SAICE GCC and National Treasury GCC, Ullr produced a probabilistic risk methodology with a quantitative risk assessment structure, and Njord produced a feasibility and business-case method for development finance work.
Every output was returned to Frigg on the fleet bus, acknowledged, and ruled on. Nothing was published or reused without that ruling.
The outcome
The sweep and tiering ran in a day. The five annexes are now standing assets: they are modular by design, so the next qualified tender in the pipeline attaches an existing methodology rather than commissioning a new one.
The honest limitation is worth stating, because it is the real lesson. Assets built against a specific bid cycle expire with that cycle. Faolan now treats the annex library as a permanent asset and the tender list as a perishable one, and refreshes them on different clocks.
What this demonstrates
Qualification is a decision problem, not a search problem. An agent fleet is well suited to the volume screening and to the first draft of specialist methodology, and it is genuinely fast at both. What it does not remove is the judgement gate: a named owner still rules on what is credible, what is client-facing, and what gets binned. That gate is where Faolan''s controls discipline lives, and it is the reason the output is defensible rather than merely quick.