Here you can find a detailed description of a full technology forecasting workflow on the GetFocus platform (Odin). This is also a typical representation of how our analysts go about a client project.

For the sake of discussing a workflow with a real-world example, we take the example of 'electric vehicle battery chemistries'.

Typically you start with a broad and high-level question with several underlying conditions.

Example: We work in Electric Vehicle powertrains and want to know which technology will be dominant in this field in the future.

Break down this question into smaller questions.

In the end, you might end up with a question like: "Which emerging high energy density battery chemistries for next-gen electric vehicles are being developed?" This question will serve as the input for our technology scouting features.



Step 1 in creating a comprehensive technology forecast is always to identify which emerging technologies are competing in a given space. For this, we need to scout technologies. Scouting involves finding all related technologies in your field of choice.

We use the "Technology Scouting" and/Or "Emerging Technology Scouting" tabs in the Chat section for this purpose.

The "Technology Scouting" tab gives results for all the technologies in a given area for the search query.

The "Emerging Technology Scouting" tab primarily gives results for emerging technologies, generally with a lower technology readiness level (TRL) which may not have been (fully) commercialized yet.

The search query for "Technology Scouting" and "Emerging Technology Scouting" can be as simple as: "battery chemistries" OR with added context "battery chemistries for electric vehicles", OR with even more added context such as in our example at the top of this article "emerging high energy density battery chemistries for next-gen electric vehicles".

** Note It is important to note that when conducting a scouting search, using less context in the query will yield more results. This could be beneficial as it includes technologies that may not have been explored for the particular use case, yet hold potential for the future. Refer to the screenshots below to see the difference.



After scouting, we have a list of competing (emerging) technologies for a certain use case. Now, we have to evaluate these technologies. The primary objective of this step is to assess the suitability of our candidate technologies for the specific use case we have in mind. In this particular example, the use case pertains to battery chemistry for electric vehicles.

Evaluation entails the assessment of technologies based on various criteria commonly employed for comparing technologies within a specific domain. The AI offers automatic suggestions for these assessment criteria, but you also have the choice to define additional criteria if desired.

We use the "Technology Comparison" tab for this purpose.



In the search query of the "Technology Comparison" tab, you can type in the technology(s) that you want to compare, the AI may add some more relevant technologies to the list and produce results as shown in the example below.



If there are many technologies, we typically select the technologies that have a higher average score to proceed to the next step where we map out the associated patent landscapes.

** Note In general, if the list of technologies provided is excessively lengthy or if there are similar technologies mentioned with different names, the AI may consolidate the list. However, you have the option to write a follow-on instruction in the search query to prevent this consolidation.

After scouting and evaluation, we have reduced the list of all possibly interesting (emerging) technologies to those that score the best for our use case. Next, we need to map out the patent landscapes associated with these technologies to identify their improvement rates. In this step, we take the highest-scoring technologies/ suitable technologies for the use case to map out their patent landscapes and obtain technology improvement rates.

For each technology, we search for the patent landscape using the "Smart Search" button. The smart search will reformulate your search query into a longer and more technical description of your input. The longer and more technical reformulation aids the AI search model in retrieving the most relevant results.

After performing a smart search, you will likely find a large number of patents. Our AI search model helps you get in the right direction, but you still need to filter your results to ensure an accurate dataset. This is a crucial step! If you do not filter your dataset, the improvement rate may be calculated using inventions that do not directly relate to your field of interest, and may thus be unreliable. The steps below help create a good representative patent landscape to derive a precise improvement rate for the technology.

(Learn more about how to filter search results in How to filter search results?)



3. Sorting the patent set in the ascending order of relevance.

We do this to verify whether the patents with the least relevance in the landscape still relate to the given technology. If this is the case, you can generally be confident that more similar results also relate to your technology area of interest.

4. In the rare case when none of the patents on the first page of ascending relevance relate to the intended technology field, you can:

AND/OR

5. Once we have a patent landscape that accurately reflects the intended technology, save the set (Read how this works here Searching with Odin).

6. We repeat these steps for other technologies we want to compare such that we can compare their Technology Improvement Rates.

Introduction 

If we want to make future-proof decisions, we have to understand how quickly technologies are getting better, which is what we are measuring with the Improvement Rate.

In our research collaboration with MIT, we found that the rate of improvement of technologies can be predicted based on metrics in patent data. Crucially, we also found that the fastest-improving technologies based on these metrics typically become the dominant market solution. Here is how this method works:

How to Interpret Technology Improvement Rates (TIR):

** Note, TIR only gives us an overview of how these technologies improve today, but does NOT tell us when a given will win or dominate in the future. To predict when an emerging technology will become dominant, we use the Forecasting tool.

Interpreting the graph curves for Technology Improvement Rates:

Here we predict when competing technologies will probably dominate in the future. Please note that this feature requires you to input starting values of current cost and performance associated with the candidate technologies. This step can be tricky, as such information is not always readily available. The Forecasting tool will create charts based on any input, so be careful that you put in, garbage in = garbage out.

Here is how the Forecasting tool works:

Below is an example of a forecasting exercise.

** Note Please note this is only an example, the improvement rates and costs in the example are arbitrary values selected for a simple explanation.

Interpreting the forecasting curve:

With a current improvement rate of 67.50%, Graphene batteries, despite having a very high current cost of $1000 per kWh of energy storage, will overtake Solid-state lithium batteries by 2033.

Sodium-ion batteries will remain the most cost-effective technology for the foreseeable future, despite a lower improvement rate, as their starting costs per kWh are significantly lower.



Once you have found the best technology, you can learn quickly and in great detail about the technology by chatting with individual patents OR even a set of patents.

Summarize any patent with a single click on the "AI summary" button.

It extracts information from the entire patent text and summarizes it in a simple, easy-to-read way. The following information is displayed:



The "Chat with Invention" function lets you chat with a particular patent using natural language through a Large Language Model (LLM). The answers are ONLY derived from the given patent text. The entire patent text is taken into consideration to answer your questions i.e. all sections- Title, Abstract, Claims, Description, Publication date/geographical coverage/ assignee, etc.

Here are some examples of the most frequent use cases of "Chat with Invention" by our analysts.

The "chat with set" function lets you analyze an entire set of patents rather than individual ones using natural language through a Large Language Model (LLM). You can ask your dataset any question, the LLM will analyze it and provide you with answers.

*There are important factors to consider when using the "chat with set" feature. For more information, please refer to the following link: Chat with set .

Typical use cases for "chat with set"

Tracking

Click on "New this month" to see the latest inventions.

Ask: "Provide a summary of the inventions related to Sodium-ion batteries for this month."

Finding the Latest Trends

You can ask:

Organization and Portfolio Analysis

Comparing Inventions