Lambda Function and up2parts: How Two Founders Automated the Most Painful Part of Manufacturing Sales

Company Profiles

Lambda Function was founded by Tanmay Aggarwal in 2020 to solve a problem he had seen from inside manufacturing operations: CNC machines generate enormous amounts of data — spindle loads, vibration, temperature, feed rates, cycle times — and almost none of it is usable by managers or process engineers in real time. The data stays in the controller, or gets dumped to a file that nobody looks at. Lambda Function builds the AI-driven data pipeline layer that turns raw machine sensor output into production insights: cycle time variance, tool wear indicators, quality risk flags, and capacity utilization reporting.

up2parts (OptoParts) was founded by Marco Bauer in 2020, bringing 10 years of manufacturing industry experience to a specific problem: the CNC machining quotation process. Getting a price for a machined part is currently almost entirely manual. A customer sends a 3D CAD file. An applications engineer imports it into their CAD system, analyzes the geometry for setup requirements, selects cutting strategies, estimates cycle time, calculates material cost, adds overhead, and sends a quote — typically 2–5 days later. up2parts automates this process end-to-end: CAD file in, quote out, in minutes.

Both companies target the same market segment — precision CNC machining shops, typically 10–200 employees — and both were founded by people who came from inside manufacturing rather than from software backgrounds. That practitioner origin is not incidental. It is the reason both companies built solutions that address the actual workflow rather than a generalized version of it.

---

The Challenge

Why CNC Quoting Is So Hard

CNC quoting combines geometric analysis, process planning, cost estimating, and commercial judgment in a workflow that is currently done entirely by experienced humans. An applications engineer receiving a new part CAD file must:

• Analyze the geometry for setup requirements: how many setups, what fixturing, what tool changes
• Identify any challenging features: deep pockets, thin walls, tight tolerances, exotic materials
• Select cutting strategies and estimate cycle time for each operation
• Calculate material cost based on stock size requirements
• Apply shop-specific overhead and margin rates
• Produce a formatted quote document

For a shop receiving 50–200 quote requests per week — common for precision subcontractors serving aerospace, medical, and defense customers — this process consumes a significant portion of senior engineering capacity. Slower quote turnaround means fewer quotes submitted, which caps revenue. Errors in quotes mean either lost margin (quotes too low) or lost orders (quotes too high).

The process is currently slow enough that many shops de facto triage: they selectively quote only the jobs they are already confident about, ignoring opportunities that would require more analysis time than they can afford. AI-driven quoting removes that constraint.

Why Machine Data Is Hard to Use

CNC machines have sensors. Modern controllers from Fanuc, Siemens, and Heidenhain capture spindle current, vibration, temperature, and axis loads at high frequency. The data is there. What does not exist — in most shops — is:

• A pipeline that extracts that data and stores it in a queryable format
• Analysis that identifies which patterns indicate tool wear, fixture issues, or quality risk
• Reporting that surfaces actionable insights to managers and process engineers without requiring them to be machine data experts

Lambda Function's product is that pipeline and analysis layer. It connects to existing CNC controllers without requiring machine replacement, extracts the sensor data stream, processes it through AI models trained on machining physics and failure patterns, and surfaces actionable alerts and reports.

---

What Lambda Function Built

Lambda Function's platform operates in three layers:

Data collection: Connectivity modules for Fanuc, Siemens, and Heidenhain controllers extract sensor data from the machine's existing diagnostic channels. No hardware modification is required. Data flows into Lambda's cloud analytics platform (or on-premise, for facilities with data sovereignty requirements).

AI analysis: Models trained on large machining datasets identify patterns associated with tool wear progression, spindle bearing degradation, fixture loosening, and process capability drift. The models combine physics-based understanding of machining mechanics with statistical pattern recognition from real production data.

Actionable output: Rather than dashboards of raw data, Lambda outputs: tool change recommendations (before the tool fails rather than after), maintenance alerts (before unplanned downtime), quality risk flags (when process parameters drift outside capability limits), and production reports (actual vs. planned cycle times, utilization, downtime breakdown).

The commercial model is subscription-based per machine, with typical payback periods of 3–6 months at shops that previously had significant unplanned downtime or high tool breakage costs.

---

What up2parts Built

up2parts' quoting automation handles the full quotation workflow:

Geometry analysis: The platform accepts standard 3D CAD formats (STEP, IGES, STL) and automatically analyzes part geometry for machining requirements — feature recognition, setup count estimation, tool access analysis, and tolerance extraction from manufacturing drawings.

Process planning: AI-driven process planning selects cutting strategies and estimates cycle times based on material, required tolerances, surface finish specifications, and the shop's machine inventory and tooling library.

Cost calculation: Material cost (stock size estimation × material price), cycle time × hourly machine rate, setup time, and overhead application are calculated automatically using the shop's configured cost model.

Quote generation: A formatted quote document is produced, including delivery lead time estimation based on current queue depth.

The result: quote turnaround drops from 2–5 days to under 2 hours for standard geometries. For shops with automated CNC programming (CAM), up2parts integrates to pull cycle time estimates directly from the CAM output rather than estimating.

---

Results

Lambda Function outcomes:

• Shops deploying Lambda's tool wear monitoring report 30–50% reduction in unplanned tool breakage events, by catching wear progression before failure
• Unplanned downtime reduction of 20–35% where the primary failure modes are tool- or spindle-related
• Cycle time variance analysis has identified systematic process capability issues at several customers that were generating intermittent quality escapes — issues that were previously attributed to random variation

up2parts outcomes:

• Quote turnaround for standard geometries: from 2–5 days to under 2 hours
• Quote volume: shops deploying up2parts report 40–60% increase in quotes submitted per week, with the same engineering headcount
• Quote accuracy: automated quotes match experienced engineer quotes within 5–8% for standard geometries, reducing the margin protection that shops previously built in for uncertainty
• Order win rate: several customers report improved win rate because faster quote response time is itself a competitive advantage with buyers who are parallelizing supplier evaluations

---

Lessons Learned

1. Domain specificity beats generality for practitioner tools. Both companies deliberately built for CNC machining, not "manufacturing" broadly. The specificity enables accuracy that general manufacturing platforms cannot achieve.

2. The quoting bottleneck caps growth more than capacity does. Many precision manufacturing shops have available machine capacity that they cannot fill because they cannot quote fast enough to convert opportunities. Removing the quoting bottleneck unlocks growth that was already possible but inaccessible.

3. Practitioner founders build the right thing the first time. Bauer's 10 years of manufacturing background meant up2parts didn't need to discover what the painful problem was — he knew it. The first version of the product addressed the actual workflow rather than a generalized version of it.

4. Machine data payback is in unplanned downtime, not dashboards. The commercial case for Lambda Function is not better visibility — it is fewer unplanned stops. Shops that track unplanned downtime cost can calculate Lambda's ROI directly.

5. Trust is built through accuracy, not through promises. Both companies earn adoption by being accurate: up2parts quotes that match what an experienced engineer would have quoted, Lambda alerts that turn out to be real problems. The AI earns authority by being right, not by being present.

---

Implementation Advice

For precision manufacturing shops evaluating AI:

If your constraint is quoting throughput — you are declining requests or slow-quoting because you can't keep up — up2parts addresses that directly. The prerequisite is a clean CAD intake process and a configured cost model.

If your constraint is unplanned downtime or quality escapes — you are losing production time to tool failures or catching defects too late — Lambda Function addresses that. The prerequisite is CNC controllers with accessible diagnostic channels (most modern controllers qualify) and sufficient production volume to make the subscription economics work.

Both tools are designed for shops that do not have large IT teams. Deployment is days, not months.

---

About the Source

This case study is drawn from AI Across the Product Lifecycle Episode 12, a podcast conversation with Tanmay Aggarwal (CEO, Lambda Function) and Marco Bauer (CEO, up2parts). See also: [[CNC Machining PLM]], [[Manufacturing Execution Systems]], [[AI in Manufacturing]], [[SMB PLM Guide]].