From Manual SAM.gov Searches to an AI Bidding Pipeline: How Loop8 Helped Jarosz Logistics Compete on Federal Contracts at SMB Scale
Jarosz Logistics partnered with Loop8 LLC to design and deliver an AI-powered federal contract bidding assistant on AWS — cutting staff time spent on opportunity discovery by approximately 90%, producing initial proposal drafts in under 30 seconds, and giving leadership a single dashboard view of the SAM.gov pipeline.
Less staff time on discovery
Automated SAM.gov ingestion replaces manual searching
Proposal drafts
Initial draft in seconds vs. days of manual writing
AI scoring accuracy
Of high-scored contracts confirmed relevant by reviewers
End-to-end flow
From contract list to generated proposal draft
Executive Summary
Federal contract acquisition is a high-effort, time-bounded process. SAM.gov publishes thousands of opportunities continuously, the response windows are short, and the proposal-writing effort per opportunity is significant — making it difficult for an SMB logistics company to evaluate every relevant opportunity and put together competitive bids on the ones that matter most. Jarosz Logistics needed a way to convert the manual, staff-intensive parts of this workflow into an automated pipeline that surfaces the right opportunities and accelerates the response.
Loop8 LLC, an AWS Consulting Partner, designed and delivered an AI-powered bidding assistant on Amazon Web Services. The engagement was structured as a four-week proof of concept that validated the core pipeline — automated SAM.gov ingestion, AI-powered opportunity matching against Jarosz's capability profile, and AI-drafted proposal response sections — followed by a production deployment that scaled the same architecture and added Amazon Textract for RFP document parsing and Amazon Aurora PostgreSQL for complex queries and reporting. The result: Jarosz Logistics evaluates a continuously refreshed pipeline of relevant federal contracts through a single operations dashboard, generates initial proposal drafts in seconds rather than days, and has redirected staff time from manual SAM.gov searching to closing the opportunities that matter.
About the Customer
Jarosz Logistics is a logistics and freight services company that pursues federal government contracts on SAM.gov as a meaningful part of its growth pipeline. The contract acquisition workflow — finding opportunities, evaluating fit, and drafting responses — had traditionally required staff time that scaled linearly with the number of opportunities pursued.
Background
The federal contract opportunity pipeline runs through SAM.gov, which publishes thousands of opportunities continuously across the federal government. Identifying logistics-relevant opportunities required staff to search the system daily — filtering by NAICS code, scanning posted opportunities, opening each promising listing, and copying key details into internal tracking. Even with experienced staff, the search effort consumed hours per week and the team knew opportunities were slipping through the cracks during busy periods.
Once an opportunity was found, the team had to make a fit judgment — does this contract match what Jarosz Logistics can deliver? That decision rested on individual reviewers comparing free-form opportunity descriptions against an implicit understanding of the company's capability profile. The judgment process was hard to scale, hard to make consistent across reviewers, and hard to audit. Opportunities, scores, decisions, and outcomes lived in spreadsheets, email threads, and individual reviewers' heads — there was no operational dashboard giving leadership a view of the pipeline.
The Challenge
Once an opportunity passed the fit screen, drafting a competitive proposal response required significant writing effort — Executive Summary, Technical Approach, Past Performance, Qualifications, and adjacent sections all needed to be written from scratch or assembled from prior responses. The blank-page problem at the start of each proposal was a real bottleneck, and the response windows were too short to handle every relevant opportunity in parallel.
Federal contracts post with strict response deadlines. SMB bidders without a dedicated proposal team are at a structural disadvantage relative to larger competitors — not because their offering is weaker, but because the time-cost of putting together a competitive response per opportunity is higher for them. Jarosz needed a way to compress the time to first draft so they could realistically pursue more opportunities within the same response windows.
“Federal contracts have always been a meaningful part of our growth pipeline, but the time-cost of finding the right opportunities and putting together competitive responses scaled linearly with how aggressively we wanted to bid.”— Jarosz Logistics Leadership
Why Loop8
Jarosz Logistics selected Loop8 LLC as their implementation partner for the combination of SMB focus, full-stack delivery, and proof-of-concept to production continuity. Loop8 delivered the complete solution — infrastructure-as-code, serverless ingestion pipeline, AI scoring and proposal-generation logic, operations dashboard, CI/CD wiring, and the production additions for document extraction and reporting — as a single integrated engagement rather than requiring multiple vendors.
The same Loop8 team that delivered the four-week proof of concept delivered the production version. The architecture validated in the PoC scaled up cleanly; the production additions (Textract, Aurora PostgreSQL) were planned at the PoC stage and slotted in without a rewrite. All source code, infrastructure-as-code, configurations, and operational documentation were transferred to Jarosz Logistics at production cutover — Jarosz owns the platform, with Loop8 available as the ongoing partner of choice rather than a dependency.
Why AWS
Amazon Bedrock provides managed access to the foundation models that power both opportunity scoring and proposal generation. Jarosz did not need to fine-tune a model, manage GPU capacity, or build ML expertise in-house. AWS Lambda runs the SAM.gov ingestion and scoring pipelines; Amazon EventBridge handles the daily scheduling. Amazon Textract was added in production to parse RFP document attachments. The data tier spans Amazon DynamoDB (high-write, key-lookup contract metadata) and Amazon Aurora PostgreSQL (relational reporting workloads). Amazon ECS with AWS Fargate runs the operations dashboard behind an Application Load Balancer with AWS WAF at the edge. AWS KMS customer-managed keys, AWS Secrets Manager for the SAM.gov API key, IAM least-privilege roles, and AWS CloudTrail audit logging give Jarosz the security and audit posture appropriate for federal-contracting workflows.
The Solution
Loop8 designed and delivered an end-to-end bidding pipeline — from SAM.gov ingestion to AI-scored opportunity matching to AI-drafted proposal sections — surfaced through a single operations dashboard for Jarosz leadership and reviewers.
Contract Ingestion Pipeline
A scheduled AWS Lambda function, triggered daily by Amazon EventBridge, queries the SAM.gov public API, filters opportunities by logistics-relevant NAICS codes, and persists contract metadata to Amazon DynamoDB. The SAM.gov API key is stored in AWS Secrets Manager and injected at runtime. The ingestion pipeline runs without manual intervention; failures route to CloudWatch alarms and the operations rotation.
AI-Powered Opportunity Matching
A separate AWS Lambda function reads newly ingested contracts from DynamoDB and invokes Amazon Bedrock (Anthropic Claude) to score each opportunity 1–10 against Jarosz Logistics' capability profile. The score and the reasoning behind the score are persisted back to DynamoDB so the dashboard can surface both the rank and the explanation. The validated success criterion in the PoC — and held in production — is that at least 80% of contracts scored 7 or higher are confirmed relevant by Jarosz reviewers.
“We now see every relevant opportunity, get an AI-scored fit assessment, and have an initial proposal draft in seconds. Our team's time is spent winning the bids that matter — not searching for them.”— Jarosz Logistics Leadership
RFP Document Parsing
For opportunities whose full requirements live in PDF attachments rather than the SAM.gov listing body, the production deployment uses Amazon Textract to parse RFP documents into structured text. The parsed content feeds into both the scoring and proposal-generation pipelines — the platform reasons over the full requirement set, not just the public listing summary.
Operations Dashboard
The dashboard runs as a Streamlit-based application on Amazon ECS with AWS Fargate, behind an Application Load Balancer. Jarosz operators see the ingested opportunity list, filter by score / NAICS / date, view the AI reasoning behind each score, click through to the SAM.gov source, and trigger proposal-draft generation directly from a contract detail view. AWS WAF sits in front of the load balancer with managed rules for common exploits, rate limiting, and bot mitigation.
Proposal Draft Generation
From the contract detail view, an operator triggers proposal-draft generation. A backend AWS Lambda function invokes Amazon Bedrock with the contract requirements, the parsed RFP attachments (where present), and Jarosz's capability profile, and returns four structured response sections: Executive Summary, Technical Approach, Past Performance, and Qualifications. The end-to-end flow from contract list to generated draft completes in under two minutes.
Reporting Tier
Amazon Aurora PostgreSQL was added at production for complex queries and reporting that benefit from a relational schema — pipeline velocity by NAICS code, average score by sub-category, conversion from scored-as-fit to bid-submitted, and win-rate trends. Aurora sits alongside DynamoDB; the ingestion pipeline writes high-volume metadata to DynamoDB, while the reporting tier maintains a normalized view for analytics queries.
Security and CI/CD
AWS Identity and Access Management scopes every Lambda function to the AWS API calls it actually needs at runtime. All third-party API tokens live in AWS Secrets Manager and are never committed to code, container images, or CI/CD environments. AWS KMS customer-managed keys encrypt data at rest across DynamoDB, Aurora, S3, and Secrets Manager. AWS CloudTrail captures the full API audit trail. GitHub Actions assumes an AWS role via OpenID Connect federation — no static AWS access keys exist anywhere in the repository.
Infrastructure as Code
The entire AWS environment is defined in Terraform with reusable modules covering the ingestion pipeline, AI scoring, document extraction, dashboard, reporting tier, security baseline, and CI/CD wiring. The infrastructure-as-code, application source, and runbooks were transferred to Jarosz Logistics at production cutover.
Results
The impact was immediate and measurable. SAM.gov opportunity coverage moved from staff-bandwidth-limited to continuous automated ingestion. Staff time spent on opportunity discovery fell by approximately 90%. AI scoring delivers a relevance assessment per opportunity in seconds, with at least 80% of high-scored contracts confirmed relevant by reviewers. Time to initial proposal draft compressed from days of manual writing to under 30 seconds of AI generation, and the end-to-end flow from contract list to generated draft completes in under two minutes.
Operationally, the ingestion pipeline runs daily without staff intervention; opportunities are never missed because of staff bandwidth or vacation calendars. AI scoring against a single capability profile delivers more consistent fit judgments than individual-reviewer assessment, with the reasoning persisted so reviewers can audit and override. The operations dashboard gives Jarosz leadership a real-time view of what's been ingested, scored, bid on, and won — enabling staffing and pipeline decisions backed by data rather than intuition. A meaningful share of staff time previously consumed by manual SAM.gov searching and proposal-writing first drafts is now available for higher-leverage activities — bid refinement, customer relationships, and the qualifications that win contracts on substance.
Operational Outcomes
- Staff time on discovery~90% less
- Proposal draft time< 30s
- AI scoring accuracy≥ 80%
- End-to-end flow< 2 min
Platform Foundation
- AI inferenceAmazon Bedrock
- Document parsingAmazon Textract
- Data tierDynamoDB + Aurora
- Pricing modelPay-per-use
“Loop8 took an idea we had been describing internally and turned it into a working platform in a four-week proof of concept, then scaled it to production with the document parsing and reporting we needed for daily operations.”— Jarosz Logistics Leadership
AWS Services Used
About the Partner
Loop8 LLC is an AWS Consulting Partner that delivers production-grade cloud and AI solutions to small and medium businesses. Loop8 helps SMB customers move from idea to working platform in weeks, not quarters — bringing the engineering depth of a larger consultancy to customers who do not have the time or budget for a multi-vendor, multi-quarter program.
Turn Federal Contracting Into an AI Pipeline
See how Loop8 can automate opportunity discovery and proposal drafting for your SMB.