FollowUp Local
A sales conversation analysis tool running on a local Gemma model, invoking the model through Ollama. It classifies customer stagnation states into eight stages and outputs real concerns, cited evidence and follow-up suggestions, with all data processed locally and never uploaded to the cloud.
Timeline (last 90 days)
FollowUp Local is a tool built by developer dongwei-polyred; after the user pastes a conversation, a Gemma model running locally analyzes the customer's stall stage, real concerns, evidence citations, and suggested follow-up messages.
FollowUp Local divides customer stall states into eight stages, including waiting_on_someone_else, price_hesitation, comparing_options, gone_quiet, and others.
FollowUp Local calls the Gemma 3 model through Ollama, running entirely locally with no GPU required.
FollowUp Local uses the 4B model by default, and older laptops can switch to the 1B model via an environment variable.
FollowUp Local only generates drafts for people to read and does not automatically send any messages, and the author believes human oversight is the correct design.
The author believes open weights are crucial to FollowUp Local, for reasons including: chat records contain customer privacy and local running keeps data from leaving the computer; local inference has zero marginal cost, making a one-time purchase model feasible; the 1B or 4B model can be freely chosen and fine-tuned on industry data; and even if the author stops maintaining it, the tool and model can still keep running on the user's machine.
The author envisions two commercialization approaches: one is a packaged desktop application with the model built in, paid for once and downloaded, aimed at users who do not open a terminal; the other is a vertical version for a single industry, customizing the stage names, signals, and example replies based on how customers in that industry hesitate.
The editor believes that using schema constraints plus a little prompt engineering to bring a 4B-level local model to a usable level in vertical scenarios, this combination of structured output, one-shot examples, and keyword hints may be more effective than switching to a larger model.
All Facts (8)
The editor believes that using schema constraints plus a little prompt engineering to bring a 4B-level local model to a usable level in vertical scenarios, this combination of structured output, one-shot examples, and keyword hints may be more effective than switching to a larger model.
50%UnverifiedFollowUp Local divides customer stall states into eight stages, including waiting_on_someone_else, price_hesitation, comparing_options, gone_quiet, and others.
50%UnverifiedFollowUp Local calls the Gemma 3 model through Ollama, running entirely locally with no GPU required.
50%UnverifiedFollowUp Local uses the 4B model by default, and older laptops can switch to the 1B model via an environment variable.
50%UnverifiedFollowUp Local only generates drafts for people to read and does not automatically send any messages, and the author believes human oversight is the correct design.
50%UnverifiedThe author believes open weights are crucial to FollowUp Local, for reasons including: chat records contain customer privacy and local running keeps data from leaving the computer; local inference has zero marginal cost, making a one-time purchase model feasible; the 1B or 4B model can be freely chosen and fine-tuned on industry data; and even if the author stops maintaining it, the tool and model can still keep running on the user's machine.
50%UnverifiedFollowUp Local is a tool built by developer dongwei-polyred; after the user pastes a conversation, a Gemma model running locally analyzes the customer's stall stage, real concerns, evidence citations, and suggested follow-up messages.
50%UnverifiedThe author envisions two commercialization approaches: one is a packaged desktop application with the model built in, paid for once and downloaded, aimed at users who do not open a terminal; the other is a vertical version for a single industry, customizing the stage names, signals, and example replies based on how customers in that industry hesitate.
50%