The situation
An award-winning landscape architecture studio — a lean team of roughly 8-12 people, with three senior leaders running projects while the founder carries administration and new business — came to Node8 with a business-development problem that will sound familiar to most small AEC firms.
Several years ago the studio subscribed to a paid RFP digest service. The digests arrived, but they were broad: someone still had to manually sort through pages of solicitations to find the handful worth reading closely. The cost was hard to justify, and about three years ago the firm discontinued it. Since then, opportunity discovery has run on personal networks and ad-hoc manual searches of public solicitation portals — and revenue has trended down alongside the shrinking top of funnel.
The frustrating part: the firm is good at winning. Its historical win rate on pursued RFPs sits around 18-20%, which is healthy for public-sector design procurement. The bottleneck isn’t proposal quality — it’s consistently seeing the right opportunities in time to respond well.
What “the right opportunities” means here
This is not a firm that wants every parks-department contract in a 500-mile radius. Its portfolio centers on culturally significant public projects — work where landscape, history, and community meaning intersect. That specificity is exactly what makes generic RFP feeds painful: keyword filters on “landscape architecture” surface everything from median plantings to stormwater retrofits, and the culturally resonant projects the studio exists to do are buried in the noise.
The firm also wants to widen the aperture beyond pure government solicitations — institutions, foundations, and quasi-public entities issue relevant RFQs that never hit the standard portals. Any system that only watches BidNet-style state and municipal feeds would miss part of the market.
Node8’s approach
Node8 proposed treating this as an AI analyst problem, not a subscription problem. Rather than starting with a build, the engagement opened with a no-cost exploratory phase structured around three questions:
- What has the firm actually pursued and won? The studio is sharing its project and pursuit lists from 2024 through 2026. That history becomes the ground truth for what “good fit” means — the training signal for scoring, and the honest baseline for the current win rate and pipeline volume.
- Where do opportunities actually appear? An inventory of every source the firm uses or knows about — portals, agency sites, newsletters, the old digest service’s coverage — to map what automated monitoring needs to watch.
- Does an existing tool already do this well enough? Node8 is evaluating current RFP discovery platforms against the firm’s specific focus before recommending anything custom. If an off-the-shelf tool covers 80% of the need at a sane price, that’s the recommendation. A custom AI analyst gets built only where the existing market genuinely falls short — most likely in portfolio-specific fit scoring and go/no-go drafting, which no generic platform does.
This ordering matters. The earlier introduction call also covered the firm’s experiments with ChatGPT for proposal writing — drafts that felt generic and needed heavy editing. The lesson carried into this design: generic AI output isn’t the product. AI that has been given the firm’s actual portfolio, voice, and priorities is.
What the analyst will do
The working design is a pipeline: monitor solicitation sources continuously, extract structured details from each RFP/RFQ (agency, scope, budget signals, deadlines, selection criteria, teaming requirements), score fit against the firm’s portfolio and stated priorities, and draft a short go/no-go summary — with the reasoning shown — for a principal to review. For opportunities the firm decides to pursue, the same extracted structure becomes proposal inputs: requirements checklists, evaluation-criteria mapping, and relevant past-project matches.
The full pipeline design is covered in How an AI RFP Analyst Works: From Solicitation Feed to Go/No-Go.
Where things stand
The engagement is in progress. Data gathering (project lists, source inventory) is underway, tool evaluation is next, and the pipeline described above is a design being validated — not a shipped system. What’s already clear from the exploratory work: the firm’s decision-making bottleneck is discovery and triage, the fit definition is specific enough that generic filters fail, and a human stays in the loop on every go/no-go call.
The success criteria are equally concrete. More qualified opportunities reviewed per month than the network-and-manual-search baseline produces today. Principal time spent on triage measured in minutes per opportunity, not hours per week. And over time, a win rate that holds at or above the historical 18-20% even as pursuit volume grows — because the analyst is filtering for fit, not just filling the funnel. Those numbers will be reported honestly as the pilot produces them, not projected in advance.
Work with Node8
Node8 builds AI systems for architecture, engineering, and design firms — from opportunity discovery to proposal automation — starting with an honest assessment of whether you need a custom build at all. See our AEC practice, or talk to us about your firm’s pipeline.